The chip giant has confirmed the launch of Intel oneAPI 2022, an important update of the well-known toolkits with which the company aspires to give “super powers” to developers, a qualifier that is all a nod to the good work that Intel has been doing with its developer tools since these, at the In the end, they make life easier for them.
Intel oneAPI 2022 includes more than 900 technical improvementssaid Greg Lavender, Intel’s chief technology officer, senior vice president and general manager of the Advanced Technology and Software Group. These improvements reduce development times, improve the performance of different workloads, both at the CPU and GPU level (client and server), and conform to the main industry standards. To this we must also add the availability in several languages.
Among the most important news that Intel oneAPI 2022 brings we can highlight the world’s first unified compiler which implements C ++, SYCL and Fortran, parallel data Python for CPUs and GPUs, advanced accelerator performance modeling and tuning, as well as ray tracing visualization and artificial intelligence based workload acceleration. This cross-architecture programming model allows, through unification, to improve productivity and reduce application development times based on different architectures.
According to the chip giant, also the compatibility tool has been improved Intel DPC ++ to automatically migrate 90% to 95% of CUDA code to SYCL / DPC ++, and optimizations have been introduced to enable advanced next-generation features that are already available, or will be available in the short to medium term. AVX-VNNI, present in the Intel Core Gen12, and Intel Advanced Matrix Extension (Intel AMX), integrated into Intel Xeon Sapphire Rapids and dedicated Xe graphics are some of the most interesting examples.
New features have also been introduced in ray tracing, such as the FP16 compatibility (medium precision), and development-level support has been expanded. In this sense, a greater integration of Microsoft Visual Studio Code stands out, and the compatibility with Microsoft Visual Studio 2022 and Microsoft WSL2 for the development of Linux on Windows.
Last but not least, significant performance improvements have been implemented in artificial intelligence, thanks to optimizations for TensorFlow and PyTorch. Intel’s new extension for Scikit-learn has also had a big impact, as speeds up machine learning algorithms more than 100 times (at the CPU level), and with Intel Neural Compressor it is possible to achieve higher inference performance.
